Calibrated Mineralogy Interpretation Model for Multi-Fluid Subsurface Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mineralogy interpretation techniques in the oil and gas industry produce poorly-defined, tediously manual, and ineffective multi-mineral, multi-fluid interpretation models due to the need for manual updates of end-members based on varying environmental and logging technique factors, limiting their ability to accurately predict component volumes and interpret new measurements.
Innovation Solution
A method for generating a calibrated multi-mineral, multi-fluid interpretation model by calibrating component end-members via inversion of core and specialized log data across all depths, incorporating these calibrated end-members into the model, and using them to generate component volume fraction profiles that account for uncertainty, allowing for automated generation of linear regressions and re-calibration in response to new data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual iteration is used to update end-members until interpreted component volumes match core measurements, then the interpretation model can be calibrated to match existing data, but the process becomes tedious and manual, reducing productivity
Solution Approach 1:
The patent replaces the manual mechanical process of iteratively updating end-members with an automated computational system. The computer automatically performs the inversion of well-log measurements to determine component volumes and updates end-members based on calculated statistics (mean, standard deviation) from multiple interpretations, eliminating the need for manual iteration while maintaining or improving accuracy.
Solution Approach 2:
The interpretation model performs self-calibration by automatically updating its own end-members based on statistical analysis of multiple component volume interpretations. The system uses the calculated mean and standard deviation from multiple interpretations to automatically adjust end-members, making the model self-correcting and eliminating dependency on manual intervention.
2Measurement precision
If end-members are updated manually to match core measurements, then the model fits existing data, but the model becomes poorly-defined and ineffective for predicting new measurements
Solution Approach 1:
The patent performs preliminary statistical analysis by generating multiple component volume interpretations from the same well-log measurements using different initial end-members. Before finalizing the model, the system calculates mean and standard deviation from these multiple interpretations to determine optimized end-members, ensuring the model is robust and predictive rather than merely fitting a single dataset.
Solution Approach 2:
The system implements feedback by using the results of multiple component volume interpretations to automatically update and refine end-members. The calculated statistics (mean and standard deviation) from multiple interpretations feed back into the model to improve end-member definitions, creating a continuous improvement loop that enhances predictive capability.
3Measurement precision
If end-members are customized for specific environments and logging techniques, then the model accurately represents local conditions, but the model complexity increases and requires frequent updates
Solution Approach 1:
The patent creates a universal end-member library that can be applied across different environments and logging techniques. By generating multiple interpretations with varying end-members and using statistical analysis to determine optimal values, the system develops end-members that are robust to environmental variations and logging technique differences, reducing the need for environment-specific customization while maintaining accuracy.
Solution Approach 2:
The system automatically adjusts end-member parameters based on statistical analysis of multiple interpretations. By changing end-member values systematically based on calculated mean and standard deviation from multiple interpretations, the model adapts to different conditions without requiring manual reconfiguration, reducing complexity while maintaining precision.
4Reliability
If multiple interpretations are generated and averaged to reduce uncertainty, then the reliability of component volumes improves, but the computational time and processing requirements increase
Solution Approach 1:
The patent generates multiple component volume interpretations (excessive action) but uses statistical analysis to efficiently summarize the results. By calculating mean and standard deviation from multiple interpretations and using these statistics to update end-members, the system achieves uncertainty reduction without requiring exhaustive processing of every possible interpretation, balancing reliability improvement with computational efficiency.
Data Source
AI summary
A method for calibrated multi-mineral, multi-fluid interpretation is provided herein. The method includes generating a multi-mineral, multi-fluid interpretation model for a number of log types using core and/or specialized log data acquired from subsurface region(s) that relate to components within the subsurface region(s). Generating the model includes: (1) for each log type, calibrating component end-members for the log type via an inversion of the core and/or specialized log data relating to the components across all depths of interest; and (2) incorporating the resulting calibrated end-members for the log types into the model. The method also includes generating component volume fraction profiles using log data acquired from analogous subsurface region(s) using the model, wherein the log data relate to any of the log types used to generate the model. Each component volume fraction profile includes a range of component volume fractions that accounts for a degree of uncertainty within the model.